21 Aug 2013 Confidential
On-target Rapid Prototyping using Simulink and Embedded Coder
- P. Gandhimathi (Electronics and Advanced Technologies/Research and Advanced Engineering)
2 Confidential |
Agenda
This is an example text. Go ahead and replace it
Introduction 1
2
3
4
Case study
Benefits
How do we apply PIL
5 Application
6 Features that can help us
3 Confidential |
Introduction
Efficient development with simulation and verification of possible solutions in advance.
Reduce non homogeneous behavior between simulation environment and actual control
hardware.
Target specific rapid prototyping of control system applications with MathWorks tools.
Optimization and correction of the application on-target with simulation in Simulink.
Lot of time and effort saving.
4 Confidential |
Case Study
Increased computerization of modern vehicles.
Continuous increase in number of ECUs in vehicles, with increasing safety and comfort
requirements.
Need to avoid increasing number of ECUs in vehicles.
Fitting the new applications in the Existing ECUs.
MathWorks tools support in making efficient process.
5 Confidential |
Fitting new application to existing ECU
Identifying following characteristics.
1. Interfaces
2. Memory
3. Periodicity
Challenge of identifying the Memory and Periodicity needs of the new application.
Impact of identifying memory needs at the end stage of development.
Identifying performance characteristics in simulation environment.
MATLAB Embedded Coder software in checking performance.
1. Verifying the deployment object code on target processors without modifying the original
model.
2. Memory need in target processor with .map files.
3. “Real-Time Execution Profiler” giving the timing needs on real time processor.
Working with fixed-point code when existing ECU is a fixed-point processor.
6 Confidential |
Fixed-point conversion
Providing low cost solution.
Fixed point processors requiring the fixed-point code.
Effort and error on manual fixed point calculations.
Fixed-point calculations inside MATLAB with Simulink Fixed-point Tool.
Easy comparison of results between floating-point and fixed-point model simulations.
Easy debugging and tuning with data type visible at each level.
Comparison of floating-point and fixed-point algorithm performance on target processor achieved
with PIL simulation in Simulink.
7 Confidential |
Simulation model in MATLAB IDE Real time Target Processor
Processor in loop simulation
10 Confidential |
Reuse of test cases for PIL
Test Inputs in MATLAB
Floating point Simulink model
S-Function Code Running On Processor
Result Comparison
11 Confidential |
Validation Flow – Impact with MathWorks Tools
Requirement
Application code Development
Design Test Cases
for Design
Test Cases for
Code
Code change for processor
specific
Requirement
Floating/Fixed-point model
Simulink Model
Test Cases
Code generation with
Embedded Coder
Validation
Without MATLAB With MATLAB
Validation
Validation
Validation Test Cases for
Processor
12 Confidential |
Hex file
.C .C++ REPORTS
.map file
Working with MATLAB
Iteration
Analysis & Verification
13 Confidential |
• Built-in fixed-point
operations save time in
simulation.
• Multiple simulations with
different word length and
scaling to see the simulation
results before committing to
hardware.
• Generates code for supported on-
target rapid prototyping boards.
• Code can be executed on
processors to verify behavioral
performance and gather resource
utilization metrics (Memory)
Through processor-in-the-loop and
profiling techniques.
Benefits
14 Confidential |
Application
New safety regulation needs - optional features in vehicles to become mandatory.
Cost effective and competitive approach to provide better product to our customer.
Implementing new features in to the existing ECUs.
TI processors such as C2000 and C6000 processors for some High speed calculation algorithms,
Vision based applications respectively.
Successful work with TI C2000 processors through MathWorks tools (Simulink, Fixed-point &
Embedded Coder) for Electric vehicle applications.
Verification, tuning and optimization of complex control system applications.
Time, effort and cost saving.
15 Confidential |
Features from MathWorks that can help us on PIL simulation
While continuing our work on innovative solutions to our customer,
MathWorks Tools will assist us in future too.
Expectations from MathWorks on processor in loop simulation
1. IDE support on Microsoft Windows7 for MATLAB 2011
2. Few more Embedded Target support for automotive applications
(such as Microchip, ST)